Interactions of climate change with biological invasions and land use in the Hawaiian Islands: Modeling the fate of endemic birds using a geographic information system

Tracy L. Benning*†, Dennis LaPointe‡, Carter T. Atkinson‡, and Peter M. Vitousek§

*Department of Environmental Science, Policy and Management, Ecosystem Sciences Division, University of California, Berkeley, CA 94720-3110; ‡National Biological Service, National Wildlife Health Center, Hawaii Volcanoes Field Station, P.O. Box 218, Hawaii National Park, HI 96718; and §Department of Biological Sciences, Stanford University, Stanford, CA 94305

Contributed by Peter M. Vitousek, June 21, 2002 The Hawaiian honeycreepers (Drepanidae) represent a superb transmission of malaria to native birds is significant (10). At illustration of evolutionary radiation, with a single colonization higher elevations, both Culex populations and prevalence of event giving rise to 19 extant and at least 10 extinct species Plasmodium decline while native birds reach their peak abun- [Curnutt, J. & Pimm, S. (2001) Stud. Avian Biol. 22, 15–30]. They also dance and diversity. These high-elevation forests are the last represent a dramatic example of anthropogenic extinction. Crop refuge for 10 species of endangered forest birds, 8 of which are and pasture land has replaced their forest habitat, and human honeycreepers: the Hawaii creeper (Oreomystis mana), palila introductions of predators and diseases, particularly of mosquitoes (Loxioides bailleui),`kiapola a `u a (Hemignathus munroi), Hawaii and avian malaria, has eliminated them from the remaining low- `kepaa (Loxops c. coccineus), and`lala a (Corvus hawaiiensis)ofthe and mid-elevation forests. Landscape analyses of three high- island of Hawaii, the`kohekohe a (Palmeria dolei), parrotbill elevation forest refuges show that anthropogenic climate change (Pseudonestor xanthophrys), and po o`uli (Melamprosops phaeo- is likely to combine with past land-use changes and biological soma) of Maui, and the puaiohi (Myadestes palmeri) and Kauai invasions to drive several of the remaining species to extinction, `kepaa (Loxops c. caeruleirostris) of Kauai. Several other species especially on the islands of Kauai and Hawaii. remain federally listed as endangered, but there is little hope that any individuals will survive. Other native species remain abun- ossil evidence shows that the Hawaiian Islands were once dant in these higher elevation forests, although some of them are Fhome to more than 100 endemic species and subspecies of declining as well. land and water birds (1). The arrival of Polynesians and, subse- There is substantial evidence from studies of human as well as quently, Europeans and other colonists ended the isolation that avian malaria that the development of Plasmodium parasites fostered the evolution of this diverse avifauna. Currently, 48 of within mosquitoes is temperature-dependent and that there is a the more than 100 original species are listed as extant; however, threshold temperature below which Plasmodium cannot develop 11 of these species have not been seen in more than a decade and to its infective stage (11, 12). In Hawaii, the threshold temper- are probably extinct (2). To date, extinctions of the honeycreep- ature for transmission of Plasmodium relictum has been esti- ers in particular have been driven largely by habitat loss, mated to be 13°C, whereas peak Plasmodium prevalence in wild introduced predators, and diseases (3–6). Habitat loss began mosquitoes occurs in mid-elevation forests where the mean with Polynesian colonists, who cleared much of the low-elevation ambient summer temperature is 17°C (10). and seasonally dry forest for agricultural purposes (7). Europe- Could anthropogenic climate change exacerbate the effects of ans and other colonists introduced new agricultural technology an introduced temperature-dependent disease, driving many of and additional domesticated animals; the introduction of cattle the remaining native birds over the brink to extinction? The key and sheep in particular led to the development of pasturages in climatic features of Hawaiian montane forests are the northeast high-elevation forest that further reduced the amount of suitable trade winds and the associated trade-wind inversion. The alti- habitat for native forest birds (8, 9). Today, the distribution of tude of this inversion averages 1,900 m; above that altitude, montane tropical rainforest is constrained by agriculture and humidity and precipitation decline rapidly, setting an upper limit urban development at low elevations, and pasture remains the to forest vegetation (13). Although the Hawaiian Islands are major human land use at high elevations. maritime, and so relatively well buffered climatically, palyno- The Polynesians hunted some birds for food and feathers and logical studies show that both local temperatures and the ele- also introduced rats and dogs that preyed on nesting birds. Later vation of the trade-wind inversion have responded substantially colonists introduced several additional predators, most notably to past climate changes (14–16). How will global climate change mongooses, cats, and two additional species of rats. For forest- affect the regional and local climate in Hawaiian montane dwelling native birds, honeycreepers in particular, the accidental rainforests? introduction of mosquitoes and avian pox, and more recently Although detailed predictions of regional climate remain avian malaria, has had the greatest consequences. Field research uncertain, particularly in the tropics, recent studies that focused in Hawaiian forests shows that native bird abundances, malarial on montane forests of the Neotropics provide some evidence for parasite prevalence, and mosquito vector levels vary predictably a common climatic response under a warming scenario. Based on along elevational gradients (ref. 5; Fig. 1). The introduced global climate simulations under elevated CO2 conditions, Still mosquito, Culex quinquefasciatus, is present in high numbers at et al. (17) suggest that sea-surface warming in the Pacific will lower elevations where many introduced birds and almost no cause an intensification of the tropical hydrologic cycle, which in native birds are found. Because the introduced birds are resistant turn will cause the release of latent heat after condensation, to malaria, the prevalence of Plasmodium in avian populations

from those elevations is low. Malaria prevalence increases †To whom correspondence should be addressed at the present address: Department of significantly in mid-elevation forest, corresponding to the lowest Environmental Science, Harney Science Center, Room 528, 2130 Fulton Street, University elevations at which native birds are found currently, and here of San Francisco, San Francisco, CA 94711-1080. E-mail: [email protected].

14246–14249 ͉ PNAS ͉ October 29, 2002 ͉ vol. 99 ͉ no. 22 www.pnas.org͞cgi͞doi͞10.1073͞pnas.162372399 Downloaded by guest on September 30, 2021 and͞or private owners manage all these areas as parks and͞or preserves for conservation; all support substantial research on the endangered Hawaiian honeycreepers. We assumed that a 2°C increase in regional temperatures would occur by some time late in the next century, in keeping with climate-model predictions for the region. We further assumed that the trade-wind inversion would increase in altitude, bringing increased precipitation to high-elevation forests. Finally, we assumed that the rate of temperature change and of expansion in the transmission of malaria will be substantially more rapid than the rate of forest expansion above the current treeline. Our analyses entailed projecting spatially the isotherms for the critical temperatures for Plasmodium development under both current conditions and a 2°C warming scenario. Using a 30-m U.S. Geological Survey digital elevation model for each island and a lapse rate of Ϸ6.5°C͞km (23), we mapped the location of the 17 and 13°C isotherms as a vector data layer. The area below Fig. 1. A generalized model of native bird abundances, malarial para- the 17°C isotherm (increasing temperature) is the region where site incidence, and mosquito vector levels along an elevation gradient in malarial infection is certain, and above the 13°C isotherm Hawaii [Reproduced with permission from ref. 5 (Copyright 1986, Ecol. Soc. (decreasing temperature) development of the parasite does not Am.)]. occur. Next, we overlaid the vector coverages of the refuge boundaries and temperature isotherms onto a false-color com- posite Satellite Pour l’Observation de la Terre (SPOT) satellite warming the atmosphere. From these results, Still et al. (17) and image collected in January of 1995. We then calculated the Pounds et al. (18) suggest an increase in the lifting condensation change in forest reserve area under the warmer climate scenario level and the height of orographic clouds. These changes, termed for the following three temperature zones: above 17°C, between

the lifting cloud-base hypothesis, would increase both the lower 17 and 13°C, and below 13°C. These zones correspond to a ECOLOGY and upper altitudinal limits of montane cloud forests (19). high-risk zone for malaria infection (above 17°C), a transition Changes consistent with this prediction have been observed in zone where some transmission is possible but limited (between Central American montane forests, although an alternative 17 and 13°C), and a low-risk zone (high-elevation forest at or mechanism, upwind deforestation of lowland forests, also could below 13°C), respectively. Some malarial infections are present increase convective and orographic cloud bases (20). in these low-risk zones, but they may reflect mobility of the birds. In Hawaii, the upper treeline on Haleakala, Maui increased in Transmission, if it occurs at all in the uppermost temperature elevation during the Holocene climatic optimum when postgla- zone, is likely brief, episodic, and limited to periods of warm cial radiative forcing was at its maximum (21). This observation weather (ref. 24 and C.T.A., unpublished data). is consistent with the lifting cloud-base hypothesis. Alternatively, Loope and Giambelluca (22) suggest that the El Nin˜o phase of Results and Discussion El Nin˜o͞Southern Oscillation could provide an analog for the Results for the three areas are shown in Table 1 and Fig. 2. local consequences of a warmer world. Under these conditions, Hanawi Forest on the island of Maui (Fig. 2a) yields the most higher temperatures are associated with high-elevation drought, straightforward and hopeful result in response to the climate- which ultimately would depress the limits of cloud forest. In our change scenario. For this preserve, the area of forest with a low analysis, we assumed that lifting cloud bases and increases in risk of malarial infection (below 13°C) is cut in half with 2°Cof upper-elevation treeline will predominate. The alternative of warming. Although a reduction of this magnitude is likely to upper-elevation drought and a depressed treeline would be far affect endemic forest bird populations substantially, Hanawi more damaging to the Hawaiian forests and to the honeycreepers represents the best-case example, because past land use practices that they support. did not include the clearing of high-elevation forests for pastures. In contrast, 2°C of warming nearly eliminates low-risk forest in Methods Hakalau Wildlife Refuge (Table 1; Fig. 2b). The predominant We evaluated the probable effects of climate change on the land use above Hakalau Refuge is pasture land, which constrains extent of forests with low risk for avian malaria for three the amount of forest available at higher elevations and could Hawaiian Islands: Hawaii, Maui, and Kauai. Within each island prevent migration of forests upslope. Restoration of high- we chose a specific area representative of the most intact forest elevation forests above the refuge is crucial to improving the and highest abundance of native birds: the Hakalau National chances for survival of the honeycreeper species, particularly the Wildlife Refuge on the island of Hawaii, the Hanawi Forest on Hawaii a`kepa,a cavity nester that requires large trees (25). Maui, and the Alakai Swamp region on Kauai. Federal, state, Restoration efforts underway now at these higher elevations

Table 1. Total forest area in hectares for each zone under current (before) and after 2°C warming scenario for each island refuge Hanawi Hakalau Alakai region (3,166 hectares) (12,999 hectares) (15,326 hectares)

Zones Before After Before After Before After

Above 17°C 1,266 1,995 650 5,200 0 12,937 Between 17 and 13°C 1,235 886 9,229 7,669 15,236 2,299 Below 13°C 665 285 3,120 130 0 0

Total refuge area is shown in parentheses; values are rounded to the nearest whole number.

Benning et al. PNAS ͉ October 29, 2002 ͉ vol. 99 ͉ no. 22 ͉ 14247 Downloaded by guest on September 30, 2021 Fig. 2. Projected changes in forest cover in relation to 17 (yellow) and 13°C (white) isotherms under current and 2°C warming conditions. Changes are shown for Hanawi Reserve (blue boundary) on the island of Maui (a), Hakalau Refuge (blue boundary) on Hawaii (b), and the Alakai swamp region on the island of Kauai (c). Images were created by using a 1995 Satellite Pour l’Observation de la Terre (SPOT) false-color composite image draped over 30-m digital elevation models for each island.

could significantly affect mature forest cover 100 years or so in climate change. The Alakai Swamp region on Kauai occupies the future. the top of a plateau that has been eroding for millions of years Finally, the island of Kauai offers the least hope for main- as the island itself has been subsiding. It presently supports no taining endemic honeycreepers in the face of malaria and forest above the 13°C isotherm, and no area is free of malaria.

14248 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.162372399 Benning et al. Downloaded by guest on September 30, 2021 The area above the 17°C isotherm is indicative of transition prospectively evaluating human-caused changes in climate (28– forest, where transmission is limited and the prevalence of 30). Some of the latter studies note that other human-caused Plasmodium in mosquitoes is tied to episodic warming events changes could interact with climate change to make it more (ref. 24 and C.T.A., unpublished data). Under the warming difficult for species to shift their ranges in response to climate scenario, this isotherm shifts upward Ϸ300 m in elevation, perturbations (31). Our analysis clearly illustrates this point in corresponding to an 85% reduction in transition forest area that interactions of climate change with land-use change and (Table 1; Fig. 2c). On this island, prevention of the disease in biological invasions exacerbate the direct effects of climate the remaining populations must become the main conserva- change substantially. tion focus. We thank Hawaii Volcanoes National Park and the U.S. Fish and Conclusion Wildlife Service for providing logistical and technical support. We thank Erin Bohensky and David Saah for assistance with the geographic There have been a number of analyses of species range shifts in information system analysis, and Mary Cadenasso, Gretchen Daily, Jack response to climate change, retrospectively using glacial and Ewel, and S. T. A. Pickett for providing comments on a previous version interglacial records of climate (26, 27), and currently and of this manuscript.

1. Olson, S. L. & James, H. F. (1982) Science 217, 633–635. 15. Hotchkiss, S. C. (1998) Dissertation (Univ. of Minnesota, Minneapolis). 2. Jacobi, J. D. & Atkinson, C. T. (1995) in Our Living Resources: A Report to the 16. Hotchkiss, S. C. & Juvik, J. O. (1999) Quat. Res. 52, 115–128. Nation on the Distribution, Abundance, and Health of U.S. Plants, Animals, 17. Still, C. J., Foster, P. N. & Schneider, S. (1999) Nature 398, 608–610. and Ecosystems, eds. La Roe, E. T., Farris, G. S., Puckett, C. E., Doran, P. D. 18. Pounds, A. J., Fogden, M. P. & Campbell, J. H. (1999) Nature 398, 611–614. & Mac, M. J. (U.S. Dept. of Interior, Natl. Biol. Service, Washington, DC), pp. 19. Pounds, A. J., Fogden, M. P. & Campbell, J. H. (1997) in Meeting Report, Bird 376–381. Life International͞WWF Workshop on the Impacts of Climate Change on Flora 3. Warner, R. E. (1968) Condor 70, 101–120. and Fauna, ed. Briggs, B. (R. Soc. Prot. Birds, Bedfordshire, U.K.). 4. Scott, J. M., Mountainspring, S., Ramsey, F. L. & Kepler, C. B. (1986) Stud. 20. Lawton, R. O., Nair, U. S., Pielke, R. A., Sr., & Welch, R. M. (2001) Science Avian Biol. 9, 1–431. 294, 584–587. 5. van Riper, C., III, van Riper, S. G., Goff, M. L. & Laird, M. (1986) Ecol. 21. Burney, D. A., DeCandido, R. V., Burney, L. P., Kostel-Hughes, F. N., Stafford, Monogr. 56, 327–344. T. W. & James, H. F. (1995) J. Paleolimnol. 13, 209–217. 6. Atkinson, C. T., Woods, K. L., Dusek, R. J., Sileo, L. S. & Iko, W. M. (1995) 22. Loope, L. L. & Giambelluca, T. W. (1998) Clim. Change 39, 503–517.

Parasitology 111, 559–569. 23. Juvik, J. O. & Nullet, D. A (1994) J. Appl. Meteorol. 33, 1304–1312. ECOLOGY 7. Christensen, C. C. & Kirch, P. V. (1986) Occas. Pap. B. P. Bishop Mus. 26, 24. Feldman, R. A., Freed, L. A. & Cann, R. L. (1995) Mol. Ecol. 4, 663–673. 52–80. 25. Freed, L. A., Telecky, T. M., Tyler, W. A., III, & Kjargaard, M. A. (1987) 8. Cuddihy, L. W. & Stone, C. P. (1990) in Alteration of Native Hawaiian `Elepaio 47, 79–81. Vegetation: Effects of Humans, Their Activities, and Introductions (Coop. Natl. 26. Nowak, C. L., Nowak, R. S., Tausch, R. J. & Wigand, P. E (1994) Am. J. Bot. Park Res. Stud. Unit, Univ. of Hawaii, Honolulu), pp. 1–138. 81, 265–277. 9. van Riper, C. & Scott, J. M. (2001) Stud. Avian Biol. 22, 221–233. 27. Davis, M. B. & Shaw, R. G. (2001) Science 292, 673–679. 10. LaPointe, D. A. (2000) Dissertation (Univ. of Hawaii, Manoa). 28. Parmesan, C. (1996) Nature 382, 765–766. 11. Lindsay, S. W. & Martens, W. J. M. (1998) Bull. W. H. O. 76, 33–45. 29. Wayne, P. M., Reckie, E. G. & Bazazz, F. A. (1998) Oecologia 114, 335–342. 12. Patz, J. A. & Reisen, W. K. (2001) Trends Immunol. 22, 171–172. 30. Sagarin, R. D., Barry, J. P., Gilman, S. E. & Baxter, C.H. (1999) Ecol. Monogr. 13. Juvik, S. P. & Juvik, J. O. (1998) in Atlas of Hawaii (Univ. of Hawaii Press, 69, 465–490. Honolulu), 3rd Ed. 31. Gryj, E. (1998) in Conservation Biology, eds. Fiedler, P. L. & Kareiva, P. M. 14. Gavenda, R. T. (1992) Pac. Sci. 46, 295–307. (Chapman & Hall, New York), 2nd Ed., pp. 478–496.

Benning et al. PNAS ͉ October 29, 2002 ͉ vol. 99 ͉ no. 22 ͉ 14249 Downloaded by guest on September 30, 2021